In the newly implemented Ports interface, for Windows and Linux/ALSA, there are "fake" balance controls: since the OS API doesn't provide the abstraction of balance controls, they are emulated in software. The problem is how these controls are implemented: One of the stereo channels is always left unaffected, while the other is lowered towards zero. This doesn't match the expected used experience: In hardware mixers, if the volume of one channel is lowered, the volume of the other channel is raised by the same amount, measured in a logarithmic scale. This way, the overall impression of signal strength, as judged by the human ear, isn't altered by operating the balance control. That's the way the software balance controls should be implemented, too.